DiSK: A Diffusion Model for Structured Knowledge
This addresses the problem of flexible and precise generative modeling for structured data, which is incremental as it builds on diffusion models but targets a specific bottleneck in existing methods.
The paper tackles the challenge of generative modeling for structured (dictionary-like) data by introducing DiSK, a diffusion-based architecture that handles text, categorical, and continuous numerical data, achieving state-of-the-art performance on over 15 datasets in tasks like synthesis and imputation.
Structured (dictionary-like) data presents challenges for left-to-right language models, as they can struggle with structured entities for a wide variety of reasons such as formatting and sensitivity to the order in which attributes are presented. Tabular generative models suffer from a different set of limitations such as their lack of flexibility. We introduce Diffusion Models of Structured Knowledge (DiSK) - a new architecture and training approach specialized for structured data. DiSK handles text, categorical, and continuous numerical data using a Gaussian mixture model approach, which allows for improved precision when dealing with numbers. It employs diffusion training to model relationships between properties. Experiments demonstrate DiSK's state-of-the-art performance on tabular data modeling, synthesis, and imputation on over 15 datasets across diverse domains. DiSK provides an effective inductive bias for generative modeling and manipulation of structured data. The techniques we propose could open the door to improved knowledge manipulation in future language models.